Soul searching continues among macroeconomists who failed to predict the economic crisis. One culprit behind the great macro-flub: the “Control Illusion,” in which economists are blinkered by overconfidence in computer models just because they, as Levitt recently wrote, “solve a problem that is really, really hard mathematically.” Speaking of the illusion of control: when it comes to which experts to listen to, people seem to value confidence over expertise. (HT: The Morning News)[%comments]
Expert Failure
TAGS:
and 
. You can also watch it on 
#8 –
So the economy is more important than the environment? I don’t see how the two are mutually exclusive.
What exactly are “we” trying to learn here?
All models need set of assumptions, boundary conditions. The accuracy of the model depend upon the knowledge of the author (either a scientist, an engineer or a economist, what ever they call themselves. However, the title do not grand the knowledge as we all know). With most of them fresh out of school with very little real knowledge of the world, try to fit historical data with simplified polynomial, added large set of arbitrate constants (justified as boundary conditions), add a supercomputer, here comes a model as complex as the quantum physics (the cat is dead or the cat is alive?). No body can question its validity without a supercomputer to proof its wrong. It is a fools game to teach large mass media to follow the “trend” (including major banks). The model only works with the believers. Create a group of believe by business school or workshop are part of the game… remember how much the bank hosting the workshop for “partners”? add few glass of wine, you got believers as multiply rabbits. Garbage in and garbage out.
I’m a 26 year old computer programmer who doesn’t know the first thing about economics, and as any of my friends will attest to, I saw this meltdown coming 3-5 years ago. (You can probably even still find my posts on fark about it when I still posted there.)
I frequently serve as a expert witness in litigation involving structural engineering. I can personally attest that juries will accept total garbage if it is put forth with confidence. This is not the jury’s fault. The problem is that they have no basis to “compare” two experts other than which one “appears” to know what they are talking about. For this reason, I personally advocate for at lease one “professional” juror in cases involving technical issue. However, I doubt this will ever happen in the US.
Dear Jones and Caliphilosopher;
I wish that I could say that I agree with you both- but cannot. The sciences have more in common than you think. As for mere theoretical models, no scientist worth his’ salt will be satisfied with a mere theoretical model. In science, an idea has to be correct in order to accomplish anything worthwhile. (Max Weber, 1918) Am afraid you may be in for somewhat of a surprise (if not a bit of a shock) and may need to adjust your thinking re the sciences just a bit.
Jones raises an interesting question- however–the same question that was raised in the Hawthorne studies. The answer is this, when the model/ theory is really grounded upon evidence, the predictions become merely logical extensions of the truth
Robyn Ann Goldstein, 2009
Hmm…
I would say that economic facts *are* directly reducible to mathematical facts…
…since economies themselves are indeed mathematical constructs which overlay and intertwine with the socio-cultural construct. The extent to which one can make *predictions* based on any given set of numbers is a subtler question.
The problem isn’t with the math, it’s with the assumption that we have a correct model and can know enough of the inputs with sufficient accuracy — that people actually trust such models so blindly. This is Kuhn’s oft-misunderstood argument. Scientific progress is contingent upon the fact that we understand that our models will *fail* and that we can then improve upon those failures, but that does not mean everyone will immediately understand a failure or agree about what exactly is wrong.
An incorrect prediction invalidates the model, not philosophy of modelling itself.
Prof. Goldstein –
I think one of the bigger questions here is trying to determine the criteria on what makes an idea right or not. In physics (or chemistry, biology, etc.) you can run some sort of test that attempts to disconfirm (or confirm) a hypothesis. The problem with economics is (at least as my understanding is of the enterprise) is what Mr. Jones hinted – that in attempting to test the hypothesis it actually affects what kind of evidence you get (for/against/inconclusive) for the result. From a testing standpoint, one ought to be able to hold certain variables fixed in order to get strong evidence either way. With economics, I don’t see a way that you can hold a certain variable fixed that does not affect the outcome of the experiment. To illustrate this point, I’m thinking of an article by Elliott Sober regarding the attempt to test for multiple realizability – I’ll give you the citation later.
If you are correct about scientists who are worth their salt never rely on purely mathematical models, then what are we to say of macro-economists who do such a thing? Sounds to me that they are either poor scientists or pseudo-scientists. How does one test out macro-economic hypotheses without tainting the sample that they are trying to analyze?
Blue92 –
I thought that economies themselves were social constructs – on the micro level, economies are (as far as I understand them to be) comprised of social practices, social institutions and the actions of individuals. i am not too sure that a mathematical construct (or better yet, how a mathematical construct) has had the same effect on my actions as a social construct.
If economic facts *are* directly reducible to mathematical facts, then you are going to have to overcome the following problems:
The first is that there are no bridge laws that *directly* connect economic laws to mathematical laws. As a matter of fact (no pun intended), I’m not too sure what economic laws are – do they fall under the realm of ceteris paribus laws? There will have to be a strong account of what economic laws are in order to get this worry resolved. There is an interesting article by Jerry Fodor (Rutgers University) and Elliott sober (University of Wisconsin, Madison) regarding this particular topic in the philosophy of science.
The second is that there seems to be the implicit assumption that science is unified. That project attempted to get off the ground almost a century ago with the logical positivists (especially earlier Wittgenstein – see the Tractatus for his attempt to reduce things to logical form; also see Russell and Whitehead with respect to their attempts to reduce mathematics to logic in the Principia Mathematica). Godel’s incompleteness theorem was the straw that broke the unification/reductionist camel’s back.
I wholeheartedly agree that there is a common misconception regarding Kuhn’s work. If we understand that our models are to fail, and if there is wholehearted disagreement as to what it is that made these models fail (see the Chicago School and the Harvard School of economic thought), then it’s not hard to see why so little progress has been made in macro-economics. At the very least, in other sciences, there is some sort of consensus reached by the practitioners of a particular science. With economics, there is no such consensus; this disagreement has major political and socio-cultural consequences for those who have some skin in the game.
– Caliphilosopher
Dear Caliphilosopher;
sorry for delay in responding to your interesting questions- so here goes.
1) the point is– assuming the following is true– that there are a certain number of already verified assumptions/principles/ideas upon which the science of economics is grounded (which a test would not and did not alter)–capable of producing new hypotheses (which can be tested and are being tested right now)– one should be able to make further predictions, test them— and so on–
As for macro-economicists relying on pure economic models– I would say that if you ask one- would not he or she be interested in the verification of their model in the long term and until then- they remain in the realm of theoriticians and the very notion of the science of economics is subject to doubt… I will respond the next question later today- sorry– duty calls.